← Back to Projects Archive
GCP Hackathon 2023

Hackathon 2023 Powered by GCP

Hybrid Workplace Time & Resource Forecasting System built with Facebook Prophet, GCP, and Streamlit.

Overview

Developed during the Google Cloud Platform (GCP) Hackathon 2023, this system delivers an intelligent time-series forecasting application designed to tackle post-pandemic hybrid workplace challenges. By accurately forecasting office attendance, cafeteria food catering demand, and hot-desk utilization, organizations can optimize facilities operational expenditure while enhancing employee workplace satisfaction.

Key Architectural Highlights

  • Time-Series ML Forecasting Engine: Employs Facebook Prophet models (prophet_model.json) trained on historical attendance matrices, seasonal holiday calendars, and day-of-week trends to project forward workplace resource requirements.
  • Google Cloud Platform Integration: Utilizes GCP infrastructure for scalable batch data ingestion, containerized execution via custom Docker containers, and robust pipeline automation.
  • Interactive Streamlit Dashboard: Provides a responsive, real-time command interface (app.py) allowing facility directors and floor managers to simulate sudden attendance surges, adjust prediction windows, and export demand schedules.
  • Exploratory Data Analytics (EDA): Includes comprehensive Jupyter analysis notebooks (timeseries.ipynb, eda.ipynb) detailing feature selection, autocorrelation checks, and cross-validation metrics.

Technology Stack

  • Machine Learning: Facebook Prophet, Pandas, NumPy, Scikit-Learn
  • User Interface: Streamlit, Matplotlib, Seaborn
  • Cloud Infrastructure: Google Cloud Platform (GCP), Docker
  • Repository: GitHub - hackathon2023
← Return to Projects Archive